1

Can a dataflow process use ffmpeg to process video or images and if so what would a sample workflow look like

mobcdi
  • 1,532
  • 2
  • 28
  • 49

1 Answers1

3

Yes. You can make sure the necessary binaries (together with their dependencies) are available on workers using the --filesToStage pipeline option and have a DoFn in your pipeline that shells out to that binary in order to do the processing.

jkff
  • 17,623
  • 5
  • 53
  • 85
  • `DoFn` I'm not familiar with the abbreviation or term could you expand on it a bit – mobcdi Feb 12 '16 at 10:49
  • DoFn is one of the main primitives in Dataflow pipelines, which represents a function that takes an input record and produces zero or more output records; it is described in https://cloud.google.com/dataflow/model/par-do ; you may also find it helpful to work through the introductory WordCount examples https://cloud.google.com/dataflow/examples/wordcount-example – jkff Feb 12 '16 at 19:56
  • `--filesToStage` is no longer documented at that url – Travis Webb Dec 31 '20 at 05:30